Perspective on non-invasive and non-destructive photoacoustic and photothermal applications

The photothermal conversion of light to heat is a phenomenon that an average person is experiencing, albeit usually subconsciously, almost daily, in particular, on a hot summer afternoon. Photothermal phenomena typically involve dynamic light variations that cause temperature variations and heat diffusion from illuminated hot zones to shadowed zones, where the latter is determined by the structure and thermal properties of the materials involved. One of the most common natural photothermal phenomena occurring on earth is the periodic heating and cooling of the earth’s surface by the sun, which causes temperature oscillations to roughly follow day–night and seasonal rhythms, obviously strongly affected by meteorological circumstances, such as wind, rain, shadowing by clouds, variation in optical absorption and reflection of the soil, and by solar activity. In well-controlled circumstances, the material parameters involved are optical (optical absorption coefficient, optical reflection coefficient, and quantum yield), thermal (thermal diffusivity and thermal effusivity, or alternatively heat capacity and thermal conductivity), and structural, and they can be extracted from the spatiotemporal evolution of the measured temperature, provided sufficient a priori knowledge about the optical excitation and a good match between the number of data and the number of unknowns.¹ By virtue of the natural relation between the thermal diffusion length and the thermal diffusion time, which is mediated by the thermal diffusivity of a medium and by the fact that the temperature on an experimentally accessible location is determined by heat diffusion to and from its surroundings, photothermal methods have not only proven to be successful in determining thermal material parameters,¹ but also in obtaining tomographic information.² By making use of high bandwidth measurement techniques and thanks to the intrinsically high (only diffraction limited) spatial resolution of optical excitation and detection, photothermal approaches are very adequate for characterizing very small structures, spanning a depth range from cm to nm. Another asset of photothermal methods is in the field of optical spectroscopy.^3–13 In conventional optical transmission spectroscopy, the optical absorption of a medium is determined by looking at the difference between the transmitted and incident light spectrum. In the case of very small absorption, e.g., in the field of the trace gas analysis or detection of impurities in liquids, this difference is extremely small and can easily be masked by fluctuations in the light intensity. In contrast, the magnitude of photothermal signals generated in weakly absorbing samples is proportional to the optical absorption coefficient so that the detection of photothermal signals is not hampered by (fluctuations of) a large background signal.

Interestingly, photothermal excitation does not only generate changes in temperature, but via thermal expansion, also changes in density and, thus, also displacement, stress, and refractive index. This has opened a wide range of additional ways to detect photothermal phenomena with extreme sensitivity and spatial resolution and, thus, extended substantially the versatility and sensitivity of photothermal metrology and applications, e.g., by making use of optical interferometry, the Mirage (beam deflection), and thermal lens effect, to detect small, photothermally induced changes in the refractive index. Moreover, in the case of sufficiently rapid light variations, e.g., for large intensity modulation frequencies of laser light or pulsed laser beams, the involved thermomechanical actuation also leads to the generation of acoustic waves. In case photothermal signal generation happens at the surface of a material, these acoustics waves are not only generated and travel in the bulk of the sample but also along the sample surface and in the air above the sample. The latter effect has been first observed by Alexander Graham Bell and Sumner Tainter and is referred to as the photoacoustic effect.¹⁴ The dynamic nature of heating by solar illumination of different locations on earth is also partially responsible for the generation of “photoacoustic” pressure variations¹⁵ and wind. The use of photoacoustically generated sound waves in air for material characterization purposes has been first proposed¹⁶ by Alan Rosencwaig and Alan Gersho in 1980. The initial attraction of this approach, in particular, in the field of photoacoustic gas spectroscopy^17–20 is to great extent thanks to the typically very high sensitivity of microphones. Since that period, the number of photothermal and photoacoustic (PAPT) applications has been booming, and the number of applications has gone along with profound advantages in laser (excitation) technology, optical metrology, and signal acquisition technology.

In the following, with the goal of giving a flavor of their unique and strong track record and remaining potential, a non-exhaustive selection of scientific areas in which, in addition to the areas listed above, PAPT methods have already had a great impact and/or are expected to lead to great advances with respect to the current state of the art is discussed: biomedical diagnostics, cell and tissue mechanobiology, thin film and interface characterization, characterization of the microstructure of solids, and the physics of relaxation in glass-forming liquids. For each area, a concise introduction is given by the state of the art, the related track record of PAPT methods, the remaining metrological needs, and a perspective on the further potential of PAPT to address remaining characterization challenges in a non-invasive and non-destructive way.

Optoacoustic tomography is probably the application of photoacoustics that has had the most impact in science in general²¹ and biomedical research,^22–27 in particular. Optoacoustic tomography makes use of the photoacoustic effect to obtain 2D images of a sample by detecting acoustic waves that are generated by optical absorption of dynamically modulated (often, but not always,²⁷ pulsed) radiation by the sample. By performing a scan of the light sheet through the sample, stacks of 2D images are combined into a 3D tomographic data cube. Although the signal generation mechanism is different, the detection of acoustic waves by a transducer array or a matrix and the conversion of the signals to an image (e.g., by delay and sum or time reversal algorithms) resemble quite well the data processing that is used in ultrasound echography. Differences in times of arrival between locations in the sample and different detectors are exploited to obtain spatially resolved information. However, due to the different acoustic wave generation process, which involves optical absorption and thermal expansion, the information in optoacoustic signals is different from the one in ultrasound echography signals. While the image contrast in ultrasound echography mainly relies on heterogeneities in acoustic impedance, a contrast in optoacoustic images reflects differences in the optical absorption coefficient and the Grüneisen coefficient. In biomedical imaging applications, the access to information on the optical absorption coefficient with very high spatial resolution²⁷ has led to a plethora of anatomical and functional imaging applications, especially by exploiting the large optical contrast between blood and biological tissue and the substantial difference in the optical absorption spectrum of hemoglobin in blood before and after the metabolic activity in tissues.

For biomedical applications, the main limitation of the approach lies probably in the finite penetration depth of light in the human body. Currently, this is partially overcome by choosing the wavelength in the so-called biological window of the spectrum, between 600 and 1500 nm, and longer than 700 nm in case absorption by blood vessels is to be avoided.

In combination with the use of dyes and functionalized nanoparticles (with an optical absorption spectrum that strongly contrasts with the one of the tissue to be probed) administered to the patient, single wavelength and multi-wavelength (“multispectral”) optoacoustic tomography (MSOT) has shown to be very successful in localizing tumors.²⁸ 

By proper choice of the nanoparticles, namely, by synthesizing them so that their optical absorption spectrum is influenced by their surroundings, MSOT has the potential for reading out the metabolic activity, tissue composition, and even tissue temperature (for hyperthermia and cryothermia treatment monitoring^29–32), thus using nanoparticles as nanoprobes. Figure 1 shows that photoacoustic thermothermography can also be employed without exogeneous thermomarkers, by exploiting the temperature dependence of the color and composition of tissue, where the latter determine the spectrum of the optical absorption coefficient and thus of the photoacoustic signals.

There is also a perspective for use of MSOT in the field of radiation dosimetry,^33–36 which is, e.g., crucial in the proton irradiation therapy. Dyes, nanoparticles, and nanodroplets can be synthesized so that their optical absorption spectrum changes when being irradiated, making it straightforward to map the dose accumulation vs time, maximizing the dose delivery in the target region and minimizing radiation damage in the surrounding healthy tissue.

The versatility of biomedical optoacoustic tomography could also be further enhanced if the ultrasound detection would be remote, without any mechanical contact with the patient. Optical vibrometry were already well established in non-destructive testing of materials and structures, and several schemes exist for full-field imaging of vibrations, even on optically rough surfaces. Optical fiber detection in the liquid surrounding the sample^33–36 has been accomplished, and steps for replacing the piezoelectric array of ultrasound transducers with a Fabry–Pérot film³⁷ or a bundle of fibers³⁷ have been made, albeit with a required mechanical contact between the optical fibers and the skin. The challenge is to enhance the performance of non-contact optical vibrometry with respect to signal to noise (S/N) and bandwidth. Scanning optical vibrometry offers high S/N, but scanning along a large number of detection positions in order to have a high ultrasound image resolution is time consuming, which is particularly problematic due to possible movements. Full field optical vibrometry requires stroboscopic detection for a very large amount of repeated, and, thus, again time consuming, image acquisitions for different excitation pulses—probe pulse delay times. In ultrasound echography, these challenges are overcome by making use of fast electronics and parallel acquisition channels. The same concepts can in principle be used for photoacoustic imaging, provided the S/N of optical vibrometry can be pushed to the level of piezoelectric detection.

In biomedical applications and in biology research, elastic properties of tissues and cells can supply crucial information for medical diagnostics and new scientific insights. In ultrasound echography, image contrast is provided by differences in acoustic impedance, and thus in elastic modulus, between different tissue types. Differences in elasticity can also indicate malignancy of the tissue. Also in the field of cell mechanobiology and tissue engineering, access the mechanical properties of cells is of high interest to get insight into the mechanisms behind cell motility, cell signaling, cell growth, cell proliferation, cell metabolism, and effects on cell metabolism in their environment (pericellular and extracellular matrix). The spatial resolution offered by classical ultrasound echography is too low to resolve details at the cell level. Atomic force microscopy, indentation, and magnetic and optical tweezer approaches give very useful information on the elastic behavior of cells, but they do not offer 3D information. Optical coherence tomography (OCT)³⁸ and confocal optical microscopy do give 3D images of cells but with poor optical contrast, so fluorescent labels need to be used to resolve intracellular components, e.g., in the framework of getting insight into intracellular transport mechanisms.

Picosecond laser ultrasound, in which acoustic waves are photoacoustically generated by shining pulsed laser light on an optically absorbing layer on a substrate that is in mechanical contact with the sample and which are optically detected by a probe laser beam, has been convincingly shown to overcome many of the above-mentioned limitations.^39–48 The high bandwidth provided by picosecond and femtosecond laser pulses makes it possible to generate acoustic wavelengths that are shorter than optical wavelengths. The generated ultrasound wavefront, which is traveling both through the sample and through the substrate, can be detected either (i) at the photoacoustic actuator layer, e.g., by detecting the ultrasound modulated optical reflection, or, (ii) in case the sample is sufficiently translucent for probe laser light, via Brillouin oscillations⁴⁹ caused by dynamic optical interference between a static part of reflected probe beam laser light (e.g., the fraction of the probe light that is reflected at the metal layer or at a cover window) and a part that is optically reflected at the traveling acoustic wavefront due to the accompanying refractive index changes. The faster the wavefront moves, the faster the optical phase difference between the interfering beam changes, and the higher the frequency of the intensity oscillation. Both probing approaches have been shown to be very successful to obtain mechanical impedance information of cells at their interface with the substrate and anatomical images with contrast based on mechanical differences, such as, in the case of Brillouin oscillation-based imaging, differences in the speed of sound and, thus, oscillation frequency, between different regions in a biological cell. We note that Brillouin microscopy,⁵⁰ in which spontaneous Brillouin scattering of photons by acoustic phonons is detected, has a similar resolution and bandwidth as Brillouin oscillation-based laser ultrasonics. However, e.g., by making use of optical gratings in transient grating-based stimulated Brillouin scattering approaches,⁵¹ it is possible to channel a maximum of deposited energy into a wavelength range (and, thus, frequency range) of interest.

Several further advances in the application of this approach can be expected. Transient grating excitation can be used to obtain viscoelastic information inside cells and tissues with micrometer resolution (Fig. 2), and if issues of radiation damage can be overcome, there is a potential for much shorter wavelengths.^52,53 There are also possibilities to photoacoustically generate and optically detect acoustic shear waves,^54–62 which can be expected to be more sensitive to the intracellular structure than longitudinal waves. The very high bandwidth of picosecond laser ultrasound enables monitoring resonant mechanical oscillations of nanoparticles,^63–68 which, in turn, are influenced by the thermal and viscoelastic properties of their environment. The frequency and decay time of the oscillations can, thus, be used to use dispersed nanoparticles as local nanoprobes and obtain viscoelastic information in different locations in a cell. The nanoparticles can be functionalized so as to reside in particular compartments of interest in a cell. In view of monitoring transport and other dynamic processes, there is also still room for improvement in terms of image acquisition speed by making use of optical detector arrays and parallel signal acquisition, in combination with the flexibility of asynchronous optical sampling (ASOPS) laser systems.

Acoustic waves possess the great property that they easily penetrate all kinds of materials while being very sensitive to compositional heterogeneities. This is particularly useful in case of situations in which tomographic information is required on optically opaque samples. Photoacoustic generation and optical detection of ultrasound enable remote and non-destructive probing of materials and structures, e.g., for crack detection (Fig. 3).^69–80 

The analysis of reflected, scattered, or transmitted acoustic wave signals yields information on the speed of sound and the acoustic impedance of a material, which, in turn, are related to the elastic moduli and density. As mentioned above, due to the time of travel of waves being related to the length of the travel path, spatially resolved information can be obtained, such as, e.g., in ultrasound echography and photoacoustic tomography. As discussed in the previous section, picosecond laser ultrasound can be used for 3D tomography of translucent materials by analyzing the spectrogram of Brillouin oscillations in the signal.^49,81–86 In principle, in the case of optically opaque materials, backscattered acoustic waves arriving at the surface of the material where the waves are photoacoustically excited can be optically detected via the strain-modulated optical reflectivity changes in pulse-echo configuration.⁸⁶ This concept has been successfully applied for the characterization of 1D stratified materials,⁸⁷ with a depth resolution that is only limited by the acoustic travel time during the duration of the laser pulse and the depth of the optical absorption region. For 3D tomographies of soft materials like biological cells and tissues, the pulse-echo configuration is coping with the difficulty to detect weak echoes due to the small acoustic impedance mismatches at the interfaces between different regions. In an alternative approach, surface or interface acoustic waves can be photoacoustically generated, and their propagation velocity and damping can be probed by detecting tangentially propagating guided waves for different distances between the pump and the probe beam. The spatial geometry of the exciting light pattern can be varied in order to control the wavenumber and spectral content of the signal. Surface acoustic waves have the property that their probing penetration depth scales with their tangential wavelength. This property makes it possible to extract from the frequency/wavelength dependence of the velocity the depth dependence of the elastic moduli and has been applied to characterize thin films^88–101 and to perform depth profiling of different structures.^102–106 Interestingly, by monitoring the time or frequency dependence of the surface temperature after, respectively, pulsed or periodically modulated illumination, it is also possible to do photothermal depth profiling.^107–125 Just like acoustically generated displacements, thermal expansion-induced displacements can be detected optically. Alternatively, surface temperature changes can be detected via the accompanying optical reflectivity changes^126–129 and by IR radiometry^{128,130–132} or thermography. The spatial resolution and probing depth are determined by the thermal diffusion length, which can be estimated as (αt)^1/2, with α (m²/s) being the thermal diffusivity and t (s) being the time after the pulse in the case of pulse excitation, and [α/(πf)]^1/2, with f (Hz) being the light intensity modulation frequency in the case of periodically intensity modulated light excitation. Photothermal depth profiling has been demonstrated from the millimeter to the micrometer range, and an extensive toolbox of methods has been developed to solve the inverse problem, i.e., extracting tomographic images of thermal or structural properties from the time or the frequency dependence of the surface temperature.^{107–125,132} Both the photothermal and photoacoustic parts of laser-generated signals have been exploited for the determination of thermal and elastic properties of very thin films.^{92,95,97,133–139}

Thanks to the versatility and high spatial resolution of the optical generation mechanism, high accuracy and high spatial resolution have been combined to characterize the microstructure of polycrystalline materials.^140–142 Very convincing results have been obtained by using spatially resolved acoustic spectroscopy (SRAS), in which a small, fine holographically generated optical grating is used to generate monochromatic acoustic surface waves with known wavelength.^143–146 The frequency of the optically detected wave signal can be directly converted into local wave velocity, which in the case of polycrystalline materials is dependent on the crystal orientation of the local grain. In metallurgical science, the technique has been shown to be competitive with electron backscatter diffraction (EBSD), which also produces crystallographic maps of polycrystalline materials but without elastic information. The grain structure of polycrystalline materials is a determining factor for the hardness, ductility, and plastic deformation of metals, and acoustic velocity and damping values obtained from the analysis of the propagation of photoacoustically generated surface acoustic waves along longer distances are correlated to those properties, making laser ultrasonics (LUS) a suitable non-destructive method for remote monitoring of metal components in a metallurgical production chain.^143–147 

The remote character of laser ultrasonic inspection gives the method also potential for remote monitoring of the structural health of components in hostile environments⁹⁸ and on moving components.^148,149 Guided acoustic wave packets can be photoacoustically generated on one location of a moving component and detected on another location. The amplitude, time of arrival, and frequency content of the detected wave then give information on the material that the wave has encountered on its path. In case a crack defect lies along the path, then this will alter the detected signal. In moving components, cracks are often dynamically opening and closing, leading to observable modulations in the properties of a series of wave packets that are sequentially generated and detected. Such modulation can then be used as an indicator for the presence of a crack. The main challenge for the use of LUS for NDE and NDE on moving samples is probably to maintain a good signal to noise ratio and to find approaches to distinguish the wave displacements of interest from signal distortions due to other vibrations and fluctuations in the amount of reflected light. Substantial progress has already been made by dedicated approaches for optical vibrometry, e.g., speckle knife edge detection (SKED),^150,151 multiple detector laser Doppler vibrometry (e.g., Polytec-QTEC® technology), and photorefractive interferometry.^152–157 Further progress can be expected by extending those approaches to parallel array detection, allowing faster signal acquisition, row by row or image by image,^157,158 instead of point by point.

Photothermal and photoacoustic approaches have also boosted progress in the field of physical chemistry,^159–162 soft matter physics in general, and the physics or relaxation in glass forming materials, in particular.^163–172 Relaxation occurs when a part of the response of a material to a stimulus requires time consuming cooperative action.¹⁷³ For example, when a compressive force is applied to a relaxing material, then it can take a while until multiple molecular distances and molecular orientations probe different possibilities in the configurational space and adapt and form a modified network that occupies a decreased volume. This response process is characterized by a time-dependent elastic modulus. For a periodic stimulus, relaxation is revealed by a frequency-dependent magnitude of the modulus and a non-zero imaginary part. In the frequency domain, the magnitude of the modulus exhibits an S-curve, while the phase goes through a dip. The bending point of the S-curve and the maximum phase delay occurs at the so-called relaxation frequency, which is roughly the inverse of the relaxation time that characterizes the mechanical response to an impulsive force. When exciting a relaxing material with a light pulse, the material response is twofold: the temperature changes and the material expands, both the temperature evolution and the volume increase are characterized by a quasi-instantaneous response (happening at the timescale of vibrations) and a slow (relaxation) response (Fig. 4).

The sudden thermal expansion also launches acoustic waves that evacuate the induced stress away from the excited region. Also, the heat is flowing away by thermal diffusion. The response of the material, which can be monitored optically by looking at the displacement or strain-induced changes in the refractive index, is, thus, very rich, with signal features (and time-/frequency-dependent material response parameters) that reflect optical absorption-induced heating (optical absorption coefficient), change of temperature (heat capacity), thermal expansion (thermal expansion coefficient), acoustic wave propagation (elastic modulus), and thermal diffusion (thermal diffusivity). Both in the cases of transient grating excitation^{55,60,173,174} and thermal lens geometry,¹⁷³ analytical expressions are available that describe the signals of relaxing materials. The accessible frequency range by photothermally/photoacoustically induced signals is more or less determined by the spatial characteristics of the pump laser beam, with dimensions a (m) from the optical wavelength to a few 100 nm. The corresponding frequencies can be estimated as f_ac= c/a, with c (m/s) being the speed of sound, in the MHz–GHz range, for acoustic features, and f_th= a²/α, in the Hz–MHz range, for thermal features. Fluorescence-based optical thermometry has been exploited to directly measure the temperature changes that go along with the photothermal response and, thus, get unambiguous information on the frequency dependence of heat capacity. By virtue of their window on relaxation at high frequencies, the use of transient grating and thermal lens approaches has already led to interesting new insights with respect to the temperature dependence of the relaxation strength^167,171,175 and the relaxation frequency.^167,171,175 A remaining challenge is the extension of the photothermal and photoacoustic signal bandwidth to even higher frequencies and, thus, shorter wavelengths, e.g., by making use of shorter wavelength light or even x rays (e.g., from pulsed synchrotron radiation or electron laser beams)^167,171,175 and by improved stroboscopic schemes of fluorescence thermometry.^176,177 The physics of glass forming materials is also characterized by a drastic temperature dependence of the shear modulus.^55,60 Efforts are ongoing to photoacoustically generate and optically detect high-frequency shear waves.^{54–58,61,165,168,169,178–180}

The above sketched selection of the application fields of photothermal and photoacoustic approaches shows that they opened new windows of opportunity in different disciplines with high scientific, technological, and societal impact. High spatial resolution imaging and high bandwidth characterization of optical, thermal, elastic, and even rotational¹⁶⁴ properties make PAPT highly suitable for a wide range of applications involving condensed matter. With the increasing improvement in laser technology (shorter wavelengths, shorter pulses) and sensitivity and bandwidth of detection, there still remains a promising perspective for tackling new challenges in the field of non-invasive characterization of soft, biological, and heterogeneous materials.

This work was supported by KU Leuven C1 project OPTIPROBE (No. C14/16/063). The author is grateful to Lionel Larbanoix, Lei Meng, Pengfei Zhang, Liwang Liu, Peilong Yuan, and Monika Rychtarikova for their valuable contributions.

The authors have no conflicts to disclose.

The data that support the findings of this study are available from the corresponding author upon reasonable request.